Watch Dogs 2 has poor CPU optimization?

[hsjj3]

See this: http://gamegpu.com/images/stories/Test_GPU/Action/WATCH_DOGS_2/wd2_proz.png

It shows a very drastic drop in CPU performance. Unlike some other AAA games, where the CPU performance of the top CPUs are even/constant (meaning there is a GPU bottleneck), does this signify that Watch Dogs 2 has a CPU bottleneck?

I'm getting stutters on a GTX 1060 (6GB) and i5 3330 - playing on medium settings, but I don't know if I should blame the CPU or GPU for this. When I am static or indoors, it's a perfect 60fps, but when driving it is 45-50fps with frequent micro stutters.

Given how "low" the i5 6600 is on that list, and the fact that the next gen CPUs aren't going to be all that much better in terms of performance (i5 7600), added with the fact that GPU tech is just going to keep getting better and better, is it fair to say that CPUs are now going to be a limiting factor going forward? I remember 4 years ago there would be nothing that could bottleneck my measly i5 3330, but looks like buying the latest highest end i5 7600 would still lead to considerable bottleneck.

 

[lyssword]

yea

 

[AtenRa]

See this: http://gamegpu.com/images/stories/Test_GPU/Action/WATCH_DOGS_2/wd2_proz.png

It shows a very drastic drop in CPU performance. Unlike some other AAA games, where the CPU performance of the top CPUs are even/constant (meaning there is a GPU bottleneck), does this signify that Watch Dogs 2 has a CPU bottleneck?

I'm getting stutters on a GTX 1060 (6GB) and i5 3330 - playing on medium settings, but I don't know if I should blame the CPU or GPU for this. When I am static or indoors, it's a perfect 60fps, but when driving it is 45-50fps with frequent micro stutters.

Given how "low" the i5 6600 is on that list, and the fact that the next gen CPUs aren't going to be all that much better in terms of performance (i5 7600), added with the fact that GPU tech is just going to keep getting better and better, is it fair to say that CPUs are now going to be a limiting factor going forward? I remember 4 years ago there would be nothing that could bottleneck my measly i5 3330, but looks like buying the latest highest end i5 7600 would still lead to considerable bottleneck.

Upgrade to a used Core i7 3770/K, much faster in this game than any Core i5 Skylake/Kaby

 

[cbn]

Interesting that the FX-8350 beats the Core i5 6600 in that game:

(And even the FX-6300 is close to the Core i3 6100)

 

[cbn]

I'm getting stutters on a GTX 1060 (6GB) and i5 3330 - playing on medium settings, but I don't know if I should blame the CPU or GPU for this. When I am static or indoors, it's a perfect 60fps, but when driving it is 45-50fps with frequent micro stutters.

That is a beast of a GPU you have.

I would use MSI Afterburner to lower GPU clocks and see if you can create a bit of GPU bottleneck to smooths things out.

 

[LTC8K6]

Upgrade to a used Core i7 3770/K, much faster in this game than any Core i5 Skylake/Kaby

Well, we'd expect a recent i7 to do better than any i5.

I don't see any Kaby Lake testing.

Since the i5-6600 has a decent lead over the i5-4670K at a lower clock speed, I think we can expect the i5-7600 to be ahead of the Skylake i5 chips.

Would be interesting to see where Broadwell i5 and i7 would fall in this bench.

I certainly wouldn't purchase a CPU based on this game, though.

 

[frozentundra123456]

I guess whether the game is well "optimized" depends on how you look a it. You could say it is well optimized because it uses lots of cores/threads, or that it is poorly optimized because despite that it is still very demanding on the cpu. It really doesnt matter though, the performance "is what it is" no matter what. Perhaps a few patches will improve things.

I have a very similar cpu (i5 2320) and it is fine for my use because I only have a HD7770 and dont play recent demanding games (none of them really interest me that much). But the simple fact is that if one has a relatively powerful gpu and plays the latest, most demanding games, a locked i5, especially SB/IB generations, is going to struggle in a lot of games. Personally, considering the weak hardware in the consoles and how demanding recent games are on the PC, I consider most of them poorly optimized for PC, but again it doesnt really mater. The performance "is what it is" and console ports on PC require much stronger hardware, without (IMO) a corresponding increase in graphics and gameplay.

Edit: I will be interested in cpu performance charts from other review sites as well. Game.gpu used to by my go-to site for benchmarks, but recently I consider them highly questionable in a lot of cases.

 

[coffeemonster]

Interesting that the FX-8350 beats the Core i5 6600 in that game:

(And even the FX-6300 is close to the Core i3 6100)

why do those graphs always have Zambezi's like the FX-4100 but no Kaveri/Godavari or Carrizo athlons?

 

[AtenRa]

Well, we'd expect a recent i7 to do better than any i5.

I don't see any Kaby Lake testing.

Since the i5-6600 has a decent lead over the i5-4670K at a lower clock speed, I think we can expect the i5-7600 to be ahead of the Skylake i5 chips.

Would be interesting to see where Broadwell i5 and i7 would fall in this bench.

I certainly wouldn't purchase a CPU based on this game, though.

According to the OP link (pic bellow), Core i7 2600K at default is faster than Skylake Core i5 6600. Since the OP has an Ivybridge Core i5 3330 (3GHz base / 3.2GHz turbo) upgrading to an IvyBridge Core i7 3770/K will be the best vfm he/she can do to increase its performance in this game.

I dont expect Kabylake Core i5 to be faster than the Core i7 2600K not to mention the 5-10% even faster Ivy Core i7 3770K.

 

[LTC8K6]

The game likes threads overall, it seems.

So it's no surprise to see chips with more threads move up the ladder.

 

[TheELF]

Look at the CPU test video, they tested in the least CPU demanding way possible.
Anybody remember GTAV's benchmark? The last part in the inner city would give you the lowest FPS because of the amount of objects/traffic/pedestrians/action(even if it was scripted) and so on.

 

[AtenRa]

Look at the CPU test video, they tested in the least CPU demanding way possible.
Anybody remember GTAV's benchmark? The last part in the inner city would give you the lowest FPS because of the amount of objects/traffic/pedestrians/action(even if it was scripted) and so on.

Downtown

http://gamegpu.com/action-/-fps-/-tps/watch-dogs-2-test-gpu

 

[RussianSensation]

Downtown

http://gamegpu.com/action-/-fps-/-tps/watch-dogs-2-test-gpu

The i5 continues to be a disappointment in 2016. I would still recommend it with an Asrock Z170 board over the i3-7350K for those on a budget but the era of 4-core CPUs is coming to an end. If at all possible to fit into the budget, I'll start recommending Zen or i7s in 2017. Granted, this is Ubisoft we are talking about. They take console games and port them to the PC instead of developing PC games and port them to consoles like EA does. I wouldn't worry too much about upgrading my PC for Ubisoft open-world games. It's akin to throwing $$$ at a game developer who doesn't prioritize PC gaming, just to overcome lack of serious optimizations and so-so graphics most of the time.

Given how "low" the i5 6600 is on that list, and the fact that the next gen CPUs aren't going to be all that much better in terms of performance (i5 7600), added with the fact that GPU tech is just going to keep getting better and better, is it fair to say that CPUs are now going to be a limiting factor going forward? I remember 4 years ago there would be nothing that could bottleneck my measly i5 3330, but looks like buying the latest highest end i5 7600 would still lead to considerable bottleneck.

It's both.

https://www.computerbase.de/2016-11/watch-dogs-2-benchmark/3/

The best way to play Ubisoft's open-world games is 2-3 years after their release when they have been patched and next generation GPUs/CPUs have come out.

Don't forget that GTX1060 is not a true mid-range videocard but a low-end GP106 family series that used to cost $129-149. What used to be the mid-range x04 series, or $200-250 videocards have now become $400 GTX1070 and $700 GTX1080, respectively. The current GTX1080 is actually the GTX1060. The combination of Ubisoft's inability to optimize PC games and NV raising shifting every single GPU tier at least 1 level up is why it results in your GTX1060 performing poorly even at 1080p.

Change your settings from Ultra to High in the game.

 

[lyssword]

actually to answer your question it has excellent cpu utilization, as in it uses every single ounce of cpu power available, while it sucks for lower end cpu it clearly makes i7's better than i5. So glad I upgraded to xeon equal to 4770k for 220 on ebay, up from dual core g3258@4.6. Gona check this game out today, got a free game code with ssd purchase.

 

[moonbogg]

Its amazing to watch that 5960x pull so far ahead of the 6700k, despite less IPC and a 400mhz clockspeed deficit. Seeing results like this is why I'm not comfortable going to even another 6 core CPU. 8/16 or nothing for me. People so often say how games don't use more than 4 cores, and for the most part its true. But this is changing and in a pretty big way with games like this one. This is the reason why I think Zen will set the new standard in what people expect from a CPU at a reasonable price. An 8/16 Zen would wipe the floor with a 6700K in this game and without costing much more I bet.

 

[RussianSensation]

Its amazing to watch that 5960x pull so far ahead of the 6700k, despite less IPC and a 400mhz clockspeed deficit. Seeing results like this is why I'm not comfortable going to even another 6 core CPU. 8/16 or nothing for me. People so often say how games don't use more than 4 cores, and for the most part its true. But this is changing and in a pretty big way with games like this one. This is the reason why I think Zen will set the new standard in what people expect from a CPU at a reasonable price. An 8/16 Zen would wipe the floor with a 6700K in this game and without costing much more I bet.

Ya, there is a rumour that AMD may even sell a lower cocked 8C/16T Zen for $400. Let's not get carried away though as I think if AMD had 5960X level of performance, they could easily sell it for $500-600 and position it directly against Intel's $649 6850K. I have a feeling AMD's 8C/16T Zen may only be as fast as Intel's 6-core 6800K. If Zen has 40% higher IPC than Vishera, that's still a solid 25-30% behind Skylake per core. We shall see in early January. Not long to wait.

 

[Majcric]

The extra details at 100% is very demanding in this game. it can bring my i7 6700k@4.5 down into the 40's.

Anybody got a 6 or more core they can do a CPU test for this game. Very curious about the results

 

[Majcric]

Its amazing to watch that 5960x pull so far ahead of the 6700k, despite less IPC and a 400mhz clockspeed deficit. Seeing results like this is why I'm not comfortable going to even another 6 core CPU. 8/16 or nothing for me. People so often say how games don't use more than 4 cores, and for the most part its true. But this is changing and in a pretty big way with games like this one. This is the reason why I think Zen will set the new standard in what people expect from a CPU at a reasonable price. An 8/16 Zen would wipe the floor with a 6700K in this game and without costing much more I bet.

I think you would be wise to go the route you're thinking.

 

[moonbogg]

The particle effects at 100% is very demanding in this game. it can bring my i7 6700k@4.5 down into the 40's.

Anybody got a 6 or more core they can do a CPU test for this game. Very curious about the results

I have a hex core that I'd like to test the game with, but I'm actually not interested in the game enough to buy it, lol. I haven't even finished GTAV yet, so I won't get this one. But I am interested in the game from a performance analysis perspective, just not enough to actually buy the game for a few benchmarks.

 

[Psycore]

This is not 6700k this is 6700 non k.If it was 6700k should be very very close to 5960x if not better.look the clocks better,6700k doesn't come with 3.4ghz out of the box

 

[Majcric]

With further testing it appears the CPU dips is a fault of the game and not the i7 6700k. CPU usage drops in certain spots of the game world and this is when I'm seeing drops into 40's. This is not from the i7 6700k being choked out, i.e hitting a 100% and falling on its face. I would still like to see how the game runs with more core/threads thrown at it.

Keep in mind I'm driving around looking for these dips and it's not something really noticeable

 

[escrow4]

And this is why I'm still using a 5930K @ 3.7GHz. Open world games render a lot of the stuff onscreen through the CPU rather than GPU. Same with the original Watch Dogs, same with the Witcher 3.

 

[LTC8K6]

Looks like the i7-6700 is about 14% faster than the i7-2600K, so the 6700K should be well above that %.

And the i5-6600 is also about 14% faster than the i5-2500K, so the i5-6600K should also be above that %.

Then we can probably add another 8-10% for the equivalent Kaby Lake entries.

 

[Majcric]

Well the good news is for the i7 6700k owners I am not noticing any stutter. This drop happens for like a fraction of sec and the framerate then returns to 60 and beyond. Another thing this doesn't happen consistently in short intervals, you have to really go out of way to find one of these dips.

Again this is with extra details at 100% the game should scale pretty well if this setting is reduced or completely turned off. in short, more than enough frames for most gamers.

 

[ShintaiDK]

http://www.pcgameshardware.de/Watch-Dogs-2-Spiel-55550/Specials/Test-Review-Benchmark-1214553/